528
The New York Times source code leaked by a 4chan user
(stackdiary.com)
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 07 Jun 2024
528 points (98.4% liked)
Technology
59598 readers
2790 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
Thats a lot of data but surly its not all their articles cos I'd very much like to train mixtral7x8b on it along with 4chan data and shir from the dark web. Surly there is a project where such a model is public and being trained on literally everything regardless of legality.
EDIT: why am i getting downvoted?
you're getting downvoted because LLMs are simply not very good, they consume lots of energy (bad for climate), and seemingly most people involved in ai hype want to replace human creativity or something.
how about instead of training a not very trustworthy or useful LLM on lots of nyt, 4chan, and "dark web", you go read lots of nyt, 4chan, and dark web to train your own (much better) model (your brain).
They are very good they exceed the capability of many humans in many tasks. If consume energy = bad for environment then all electric vehicles are bullshit cos they have energy inefficiencies that petrol cars don't (thermodynamics is a bitch). U do realise the argument about if asking an ai to create an image is art argument is literally the same argument that was had about if photography is art.
Llm are decently trustworthy especially with chain of thought reasoning and tool capabilities. And they are extraordinarily useful people wouldnt be using them and creating a market for them of they weren't. I can't train my brain then share it for free to everyone on the internet to download I can with an ai tho.
Have you seen that study about the accuracy of chatgpt responding to programming questions? (here) It's wrong 52% of the time, and I can say that I have personally experienced trying to use chatgpt for programming and getting more confused rather than less. Maybe it is because I wasn't using gpt4, or claude, or whatever new model is the best, but I'm just sharing my experience.
Also I support electric vehicles because without them lots of energy (and emissions) is generated for critical infrastructure (we can't ditch cars yet), and so replacing that with renewably generated energy is a good idea.
LLMs consume lots of energy to train and use, but instead of literally moving millions of people around, they assist you in doing things you could have done without them, but with dubious accuracy. Look at the massive use of LLMs in by students to cheat in school, yes they may not get detected, but sometimes they have noticable flaws, that get them in large trouble for being too lazy to actually learn anything.
If you want to learn in depth knowledge about a topic, just go look it up and learn there, it's more helpful than an LLM.